RPG Evolution - The AI DM: The Downside of Free

Should you give your RPG content away?

With the arrival of Large Language Model (LLM) AI, all the content that's freely accessible on the Internet is about to dominate every topic, including tabletop role-playing games.

system-63768_960_720.jpg


Picture courtesy of Pixabay.

Free-For-All​

The nature of tabletop role-playing games has always been collaborative. Dungeons & Dragons is a machine that makes more D&D. By its very nature, game masters are expected to share the content they produce with their players.

This naturally extended to fanzines, personified by Alarums & Excursions. Fanzine distribution was initially limited, but that changed once the Internet arrived, effectively unbalancing the stranglehold then-publisher of D&D TSR had over the game's future. What was once a vehicle for D&D's popularity became an existential threat to its brand.

That tension, between fandom and capitalism, continues to this day, as Wizard of the Coast's attempts to delegitimize the Open Game License recently demonstrated.

A Middle Path​

Early role-playing game innovations were often launched by creators frustrated with perceived flaws in D&D. Over time, these games would go on to become competitors to D&D itself, Tunnels & Trolls being a prime example.

After TSR's numerous attempts to control its fandom through threats of lawsuits, it was acquired by Wizards of the Coast, who altered course. Gone were the hostile overtures towards fans, replaced by a new model in which fans could potentially make money by leveraging the core rulebooks without competing with them. The Open Game License unleashed all that fan creativity with the opportunity to make a profit. The Dungeon Masters Guild created an even closer relationship between iconic D&D settings and fans looking to share their work.

Of course, fandom has never been constrained by rules, and gaps in Internet support created an opportunity for content to flourish on sites like the D&D Wiki:
In 2015, the WotC message boards, the hotbed of creativity which spawned this very wiki, were shut down by WotC. This action occurred because it became apparent that other communities on the internet had exceeded their capacity to discuss the subject of Dungeons and Dragons effectively, and that maintaining what amounted to a link-hub to other websites was not cost-effective. It is entirely possible that D&D Wiki, along with other communities like it, may have contributed to this unfortunate event. This was followed a few months later by the opening of the Dungeon Master's Guild.
The D&D Wiki's influence cannot be underestimated:
D&D Wiki, as the largest and most active of this type of website, remains the top search result when seeking D&D homebrew on the internet, even on multiple search engines, and even outcompeting the official WotC Dungeon Master's Guild, which offers its users the opportunity to make a profit off of their work!
Similarly, Open Game License sites that publish OGL-only content are treasure troves of free RPG material, the basis of which helped launch Pathfinder and the OSR.

The tension between "for pay" sites and "for free" sites continues to this day. But neither group likely anticipated an era when robots would comb the Internet, combine its content into something human-legible, and then release it for free without citing its sources.

Feeding the Machine​

LLM AIs consume the data sets provided to them. When it comes to role-playing content, this means that tools like ChatGPT are only as good as their data sets, and those data sets are only as good as the content they have access to. Out of sheer economic necessity, LLMs pull that data from free content which is both large and accessible enough to refine the AI's responses.

In the past, releasing content for free was frequently seen as a stepping stone to monetizing it later, as Writers First explains:
There is nothing wrong with producing “free content” by writing on a blog or on social media if it’s something you enjoy. And indeed, there’s the potential to leverage that writing into economic benefit in the future.
With the advent of LLMs, there's another factor to consider. Be it a Wiki, a blog, or on social media, all that "free content" is fodder AIs.

There are now plenty of gated options that still release content at no cost, but not entirely for free. Pay What You Want is one way to ensure that a human must click through to purchase something without making it easily accessible to the Internet (and thus LLMs).

But for fans and hobbyists, releasing content for free is counterbalanced by the fact that they are feeding a machine that will neither give them credit or link back to their work.

Your Turn: Would you still give away your work for free knowing an AI might use it?
 

log in or register to remove this ad

Michael Tresca

Michael Tresca

What if the AI creates "accidental plagiarism"? Or accidental offense. For example in a story of zombie apocalypse the name of the heroine is Rafaela Herrera, from Nicaragua, and she kick-ass a lot of zombies. Then there are complains by Britishs because there was a historical character with that name in the real life, who fought against a failed English invasion in Nicaragua.

If you are using AI to create, or to complain details in your own game, with no-profits intentions, that is a different thing.
 

GMMichael

Guide of Modos
But for fans and hobbyists, releasing content for free is counterbalanced by the fact that they are feeding a machine that will neither give them credit or link back to their work.
I think a lawyer just got her wings.

Your Turn: Would you still give away your work for free knowing an AI might use it?
I . . . guess so? Putting anything on the interwebs is a risk. I've already written Lord of the Rings 2, but you don't know because it's not on the internet 🤓
 

talien

Community Supporter
What if the AI creates "accidental plagiarism"? Or accidental offense. For example in a story of zombie apocalypse the name of the heroine is Rafaela Herrera, from Nicaragua, and she kick-ass a lot of zombies. Then there are complains by Britishs because there was a historical character with that name in the real life, who fought against a failed English invasion in Nicaragua.

If you are using AI to create, or to complain details in your own game, with no-profits intentions, that is a different thing.
I don't know that's any different from a human creating accidental plagiarism. It's possible for two different people to come up with similar ideas. However, legally that will rely on a trail of evidence, which can be traced to electronic trails that humans leave behind (browser history, emails, etc.). AI has none of this, and that's the problem. There is no way to prove AI took your content and reused it, other than similarities in the final product it produces.
 

Yeah, so, maybe we all now stop feeding LLMs (or rather, training them) and stop pretending "But it's fun, and look at the cool art I made with MidJourney" is a good enough excuse? We're by now (hopefully?) all aware of at least the worst social and economical damage LLMs are already causing, after all.

(It's not so much "downside of free" but "downside of free, unregulated, tech-bro fueled, money-grabbing capitalism)
 

Stormonu

Legend
I wouldn't put stuff out in public if I didn't want someone to use it, AI or human. If an AI grabs it an incorporates it, I at least know it's getting eyes on it.
 

I wouldn't put stuff out in public if I didn't want someone to use it, AI or human. If an AI grabs it an incorporates it, I at least know it's getting eyes on it.
Yeah, just, nobody will ever know it was your work being incorporated, and the LLM won't ever admit to it either (barring leaks and stuff). Not saying that's much different from Jo Average copy-pasting your work and stubbornly pretending it's theirs, but the scale is unfathomably different - even if Jo spends every waking our pushing out other's works as his own, he won't get near what's already been sucked into LLM training data.
 

Stormonu

Legend
Yeah, just, nobody will ever know it was your work being incorporated, and the LLM won't ever admit to it either (barring leaks and stuff). Not saying that's much different from Jo Average copy-pasting your work and stubbornly pretending it's theirs, but the scale is unfathomably different - even if Jo spends every waking our pushing out other's works as his own, he won't get near what's already been sucked into LLM training data.
Sounds like irrational fear to me.
 

Would you still give away your work for free knowing an AI might use it?
Who is the end user? A human. It doesn't matter if an AI generates something derived from something I created (which is undoubtable derived from my experiences, which are often based upon what has been generated by others) which is then used by a human, or if a human generates something derived from my creation.

I worry about AI, but not because of plagiarism concerns.
 

Oncewasbenji

Explorer
This is a really interesting idea becuase it means that things released under creative commons (a very anti corporate control idea) is now inverted to being the thing most likely to feed a corporate tool.
 

Remove ads

Remove ads

Top